Where is generative artificial intelligence (AI) heading? And how does AI fit in information literacy instruction? These are just a few questions librarians Melissa Del Castillo and Hope Kelly had when conducting a recent survey of library professionals on how they are teaching about—and using—AI-powered tools like ChatGPT in instruction.
The two presented their findings on January 20 at the American Library Association’s 2024 LibLearnX conference in Baltimore. The 30-minute Ideas Xchange session, titled “ChatGPT Is a Liar and other Lessons Learned from Information Literacy Instructors,” was delivered to a standing-room-only crowd.
Del Castillo, virtual learning and outreach librarian at Florida International University in Miami, reminded attendees that ChatGPT, which launched in November 2022, is relatively new to the scene, though generative AI has been used by developers on the web for some time. But why has ChatGPT stood out?
A few reasons, she said, including its ease of use, its ubiquity, and its press attention. “When there’s a deluge of information,” she said that librarians must “try to create some type of literacy when looking at that innovation in education technology.”
But, according to Del Castillo, the reason ChatGPT is a “lying liar—there’s no way around it” is because of its reported bias (such as racial profiling and mis- and “mal-information”); use for deception (via user plagiarism and cheating); hallucinations (when AI gives nonsensical or inaccurate outputs); and privacy and copyright concerns. She cited the legal case in which many well-known authors are suing OpenAI, the maker of ChatGPT, over the alleged misuse of their work to train the chatbot.
Despite these issues, the two presenters said their survey sought to find out three things: AI’s potential benefits, how librarians are using it, and whether they are using it in instruction.
Hope Kelly, online learning librarian at Virginia Commonwealth University in Richmond, said the analysis showed that many people thought ChatGPT would improve academic productivity but were less inclined to believe it would improve learning performance.
However, many respondents “did think it’s important to bring this into their instructional workflows,” Kelly said. Their thought, she explained, was “whether I like it or not, whether I feel safe or not, this is changing our landscape and information space.”
Early conclusions from the survey indicate that while the tools were easy and useful for a lot of instructional tasks, “we’re also getting very wary of the tool due to both practical and ethical concerns,” Kelly said. “So there are lots of feelings, lots of concerns.”